A Real-Time Correction Method for Dynamic System Differential Response Considering the Uncertainty in Model Structure

The structural uncertainty in hydrological models is an important factor affecting the accuracy of hydrological forecasting, and quantifying and reducing its effect is a hot research topic. Based on the dynamic system response curve (DSRC) method, the mathematical relationship between model structural error and the change in input is deduced, and a dynamic system response curve method considering the model structure uncertainty (UNDSRC), which can distinguish the source of model structural uncertainty, is proposed. This method is applied to Dapoling and Fushui watersheds to test the comprehensive performance of the UNDSRC algorithm and to compare this algorithm with the DSRC algorithm. The results show that 1) the UNDSRC algorithm performs better than the DSRC algorithm in the actual basin tests, the average correction effect evaluation coefficients of the UNDSRC method is 0.82, of the DSRC method is 0.60; 2) the XAJ’s result is effectively corrected by the DSRC method in the two actual watersheds, and the correction effect is similar; 3) the UNDSRC algorithm is excellent and stable and can adapt to more complex underlying surface conditions, and the algorithm corrects the flood peak better than it does the runoff depth; 4) when the correction accuracy is the same, the ridge coefficient of UNDSRC method is much smaller than that of DSRC method.