Effective use of natural runoff forecast information can increase the upper limit of storage capacity of hydropower stations during flood season, make full use of hydropower resources under the premise of ensuring flood control safety, and improve the economy of power grids. For the power system with a high proportion of small hydropower resources, an optimal scheduling strategy is proposed, which takes into account the expected runoff and the upper limit of elastic storage capacity of hydropower stations within the basin during flood season. Firstly, a screening method of meteorological and hydrological forecast factors is proposed based on the maximal information coefficient (MIC), and an Informer natural runoff forecast model based on the attention mechanism is constructed. Secondly, by considering the accuracy of forecast information and the pre-discharge capacity of hydropower stations, a method of determining the upper limit of elastic storage capacity of hydropower stations based on chance-constrained optimization is proposed to excavate storage resources of hydropower stations during flood season. Finally, a case study of a small hydropower station group in Lishui, Zhejiang Province is carried out. The results show that the proposed model has an accurate forecast effect, which can improve the utilization efficiency of storage resources of small hydropower stations and reduce the system operation cost.