Abstract: An accurate and objective evaluation of its performance is important for improving and developing the new observation system. Our new-generation RDSS (round-trip drifting sounding system) innovatively surpasses the traditional sounding observation mode. It realizes the “up–drift–down” three-stage observation with a single release, which expands the ability and scope of existing sounding observations. This study examines the sensitivity of forecasts in the middle and lower reaches of the Yangtze River to the new RDSS observation data using the forecast sensitivity to observations method. Results show that assimilating conventional observations can reduce forecast errors and improve quality to varying degrees, with wind and temperature observations contributing the most significantly. The new RDSS observational data have a more significant contribution to forecasts in the target area of the middle and lower reaches of the Yangtze River. The forecast error is further reduced by 71.4% in the test period after data assimilation combined with the new RDSS data, with meridional wind and humidity observation contributing the most. Moreover, the contribution of its wind observation exhibits obvious spatial differences, with large value areas of forecast error reduction mainly distributed in the test station and its vicinity. In addition, wind, temperature, and humidity observations from the new data in the entire layer significantly positively contribute to forecast quality, while zonal wind observations in the middle and lower troposphere have a slightly negative contribution.