<p>The traditional runoff attribution reflects the variation intensity of human activities on underlying surface parameters, but it is difficult to consider the heterogeneity of underlying surface, and it is impossible to distinguish the specific influencing factors. In this study, the time-varying elastic model is used to improve the traditional attribution method, and the driving factors of human activities are divided into land, economy and industrial water use. Then, we use the traditional Budyko model, time-varying elastic model and the water energy partitioning framework to analyze the driving forces of runoff evolution and discuss them in different periods. The results showed: (1) the annual runoff was upward with an average rate of 3.54&#xa0;mm/a, and had a significant abrupt change in 1991 under urbanization. (2) The simulation effect of time-varying values is better than that of traditional model framework. The runoff driving force analysis considering urbanization is more consistent with the actual situation of the study area. (3) Comparative analysis of runoff attribution results by traditional Budyko model, time-varying Budyko model and water energy partitioning framework shows that: the attribution results of different stages are consistent, but there are slight differences in the attributions of each driving factor. (4) Precipitation and urbanization are the key influencing factors for the runoff evolution in each period of change. Consequently, urbanization development has significantly changed the precipitation-runoff response process in watershed. The results can provide some reference for disaster reduction in urbanized areas.</p>