The accurate simulation and prediction of flood response in urbanized basins remains a great challenge due to the spatial and temporal heterogeneities in land surface properties. We hereby propose an integrated modelling approach that consists of a semi-distributed conceptual hydrological model and a novel parameterization strategy. The modelling approach integrates the Xinanjiang (XAJ) model, Taihu Basin (TB) model, and Nash instantaneous unit hydrograph (IUH) into a framework. Model parameters are calibrated by optimizing their relationships with corresponding physical factors. The proposed modelling approach is applied in the Qinhuai River basin (QRB), China. The modelling approach shows satisfactory performance in flood simulation both for calibration and validation of flood events in the QRB. The approach has temporal and spatial prediction capability by using the established relationships between parameter values and physical factors. Robustness analysis reveals that the different sets of flood events used for parameter relationship calibration led to similar model performance. Numerical experiments show that impervious coverage poses strong influences on the model performance and needs to be considered in flood routing simulations for small- or medium-intensity flood events.