A torrential rainfall caused by Typhoon Goni during 23-24 August 2015 influenced some coastal areas in East China. This paper utilizes the ECMWF ensemble forecasts to study the key reason for the forecast uncertainty of the far distance torrential rainfall and uses the ensemble sensitivity method to investigate the sensitivity areas of the primary weather system. Meanwhile, the thermodynamic formation mechanisms of the torrential rainfall are also discussed. It is indicated that the predictability of ensemble forecasts of the typhoon-induced far distance torrential rainfall was obviously low and the significant forecast adjustment occurred just 24 hours before the torrential rainfall. At initial time when the systematic track error was minimum, the ensemble precipitation forecast was mostly close to the observation. But further analysis shows that the correlation between the deviation of typhoon track and the precipitation level is uncertainty. The contrastive analysis of the mid- and lower-circulations between two ensemble groups at different precipitation levels indicates that the forecast difference of the upper trough is the main cause for the forecast uncertainty. The eastward movement and intensification of the upper trough not only increased the baroclinic instability in the torrential rainfall area, but also enhanced the circulation magnitude in the side of the typhoon facing the area. The sensitivity analysis on 500 hPa geopotential height shows that, whether on the initial field or on the forecast fields, the mean precipitation in the torrential rainfall area has notable correlation with the movement and intensification of the upper trough. With the increasing forecast time, the areas with significant correlation expand downstream of the upper trough. In addition it is found that the movement of the upper trough can enhance (reduce) the intensification of the cold air on the left (right) side of the torrential rainfall area and make more moisture air on the right side of the typhoon transported to the torrential rainfall area.