Based on the WRF ETKF-3DVAR hybrid data assimilation system and 51 members of ECWMF global ensemble prediction in TIGGE data,the flow-dependent characteristics of typhoon forecasting errors,the sensitivity of coupling coefficient in hybrid data assimilation and forecast,and its effects on forecast skill are analyzed,taking typhoon Muifa for example.Results suggest that the forecasting error covariance of typhoon has significant meso-and small-scale characteristics and the structure of forecasting error covariance that is estimated according to ensemble prediction can reappear its flow-dependent nature.The optimal coupling coefficient in hybrid data assimilation scheme can better improve the qualities of analysis and forecast of typhoon than 3DVAR scheme.However,there are obvious effects on typhoon track forecast for different coupling coefficients and an improper coupling coefficient can lead to a worse result.That is,only the relative appropriate flow-dependent information in forecasting error covariance is coupled,the hybrid data assimilation scheme can have positive effects on the qualities of analysis and forecast.It shows that,in the hybrid data assimilation system,constructing a coupling weight function with adaptive ability and achieving automatically the aim to choose optimal coupling coefficient are of importance to improve forecast quality and give full play to potential advantage of hybrid data assimilation system.