Hyperspectral infrared (IR) measurements from polar orbiting satellite have been shown useful in weather forecasting and nowcasting. However, current use of hyperspectral IR measurements is limited due to massive data volume. In order for hyperspectral IR measurements to have real-time impacts on weather forecasting and nowcasting, data thinning and channel selection are the two most commonly used methods to speed up the process. Both methods possibly lose some fine scale information, which is important for meso-scale applications. This study presents a fast physical algorithm to simultaneously retrieve temperature, moisture and ozone profiles along with surface temperature and emissivity using hyperspectral radiance measurements in clear sky. By performing retrieval in Eigenvector space of radiances, the computation is about 60 times faster than before. With this technique, the AIRS and IASI sounding retrieval on single field-of-view (FOV) basis using more channels can be realized in near real-time, which further improves the capability of nowcasting. The retrieval results and applications to data assimilation will be presented.