Building renewable energy bases focusing on desert, Gobi and desert areas is a crucial component of new type power system in China. Taking into account factors such as construction costs and carbon emissions, the base has limited capacity for thermal power and energy storage, resulting in restricted flexibility. Consequently, the operation and scheduling of the base face significant challenges. This paper proposes a method of integrating day-ahead scheduling and realtime dispatch for wind-thermal-storage systems. In the day-ahead stage, the scheduling of thermal units, including on-off plans and adjustable output ranges, is determined based on a rough prediction of wind power. Then a flexibility interval is established considering the operational constraints of thermal units and energy storage. In the real-time stage, a quantile strategy is proposed based on the observed renewable outputs, eliminating the requirement for high-precision forecasts. This paper further proves that dispatch decisions generated by the quantile strategy naturally satisfy the system's operational constraints. The case study validates the effectiveness of the proposed method for the wind-thermal-storage system. The results demonstrated that the proposed method, which does not rely on point prediction, outperformed the rolling optimization method with prediction errors exceeding 10% in a 3-step scenario. It is found that the performance of scheduling can be improved by enhancing the accuracy of day-ahead forecasts or intraday short-term forecasts. The proposed method can serve as a valuable reference for the operation of large-scale renewable energy bases.