Under the "dual carbon" target, the penetration of wind, solar and other clean energy will continue to increase, producing a scheduling problem brought by the uncertainty of its power generation. Thus a two-stage distributionally robust optimization model for the wind-solar-storage system is proposed. To reduce the carbon emissions of the system, step-type carbon trading is introduced. In the model, the operating cost of the system is minimized according to the forecast information in the first stage. In the second stage, Wasserstein distance is used to construct a fuzzy set of data-driven output errors. An affine strategy is used to adjust the model so that the system can meet the worst condition in the fuzzy set and the adjustment cost is minimized. Finally, a strong duality principle is used to transform the two-stage distributionally robust optimization model into the equivalent MILP model. Through the analysis of numerical examples, it is compared and verified that the use of traditional units can be significantly reduced by considering the step-type carbon trading through reasonable scheduling, and the applicability and superiority of distributionally robust optimization model is demonstrated.