[Purposes] This paper aims to build accurate and fast wave forecasting models to ensure marine activities and beach safety. [Methods] The wave data and wind speed data in the Chinese Wave Database (CWAVE) from 2009 to 2018 were used as training samples to establish a rapid wave downscaling forecasting model that integrated a wave model with a random forest machine learning algorithm. The numerical wave model implemented calculations on a coarse grid, and wave downscaling forecasting was performed through the random forest algorithm, enabling rapid forecasting of wave elements in nearshore areas. [Findings] A long-term series forecasting of significant wave height, average period, and main wave direction for the entire year of 2019 in the offshore area of the Yangtze River Estuary is conducted. It is found that the established wave downscaling forecasting model can accurately predict the changes in typhoon waves and cold waves throughout the year. The relative error of the significant wave height calculated by the model is within 0.2% compared to the calculation results of the traditional wave model， the calculation efficiency is significantly improved, with the 96-hour short-term wave forecasting advancing from a minute to a second level. [Conclusions] The rapid wave downscaling forecasting model that integrates a wave model with the random forest algorithm can enhance the stability, accuracy, and calculation efficiency of wave forecasting, providing a new method for the operational application of wave machine learning algorithms in wave forecasting.