•A new urban hydrological model considering different land covers is proposed.•The TVGM_Urban model outperforms other two models in flood simulations.•The TVGM_Urban model has less prediction uncertainty than other two models. Rapid urbanization has great potential to adversely impact the local hydrological cycle, the environment, and ecosystems, as well as trigger severe problems, such as urban flooding, waterlogging, and water contamination. Understanding runoff generation mechanisms in urban areas helps identify the impacts of urbanization on runoff response and better simulate urban floods. However, in most urban hydrological models, urban surfaces are coarsely classified into impervious and pervious ones, which likely ignores differential rainfall-runoff responses of different urban surfaces and leads to large discrepancies between simulated and observed runoff. Here we developed the TVGM_Urban model, a new urban hydrological model based on the time variant gain model (TVGM), that represents nonlinear rainfall-runoff relationships for different urban surfaces. We applied the model to the Fenghuangcheng region of Shenzhen City, China and compared the results to those of two commonly used urban hydrological models: the Horton infiltration and Soil Conservation Service Curve Number (SCSCN) model. For each of these models, we conducted model uncertainty analysis using the GLUE method. The results showed that the TVGM_Urban model outperformed the other two models in terms of total runoff and peak flow. This could be due to the consideration of different land covers and both saturation excess and infiltration excess runoff generation in the TVGM_Urban model. In addition, the uncertainty analysis indicated better performance of the TVGM_Urban model in reducing structural uncertainty and prediction uncertainty. This study highlights the need to account for detailed land covers and different runoff generation mechanisms in urban hydrological modeling.