This study investigates the characteristics of landfalling severe typhoons (LSTYs) in South China (SC), which are defined as typhoons with a 2-minute mean maximum sustained wind ≥ 41.5 m s⁻¹. Since 1949, 13 LSTYs have been recorded in SC, with most undergoing rapid intensification before landfall. These typhoons are categorized based on the intensity of the summer monsoon: weak, moderate, and strong. LSTYs under weak monsoon conditions exhibited the smallest outer size and weakest "warm-wet" core due to a lack of abundant monsoon water vapor, and they rapidly intensified due to favorable offshore ocean warming. In contrast, LSTYs in strong monsoon conditions received poor energy supply from the coastal ocean but feed by the strong monsoon flow. In addition to analyzing the observed LSTY, we also explore an ensemble seasonal forecasting model for predicting typhoon frequency in the western North Pacific using multiple machine learning methods. Results show that mean bias of the MME for TYF is notably smaller than that of the ECMWF's most recent seasonal forecasting system (SEAS5) in the years of 2017‒2023, underscoring the potential of ensemble prediction approach utilizing multiple machine learning algorithms to improve the forecasting skill of TYF over the western North Pacific.