Daytime Total Precipitable Water from 0.91um

The next generation Geostationary Extended Orbit (GeoXO) Imager (GXI) will have a new water vapor spectral band 0.91µm. This band has a good daytime sensitivity to water vapor absorption in the atmosphere, making it capable to retrieve the total precipitable water (TPW) in the atmosphere through combined usage with observations from atmospheric window channels such as the one centered at 0.865µm. The capability of this spectral band for total precipitable water retrieval has been demonstrated before with MODIS, which provides 4 times a day from Terra and Aqua satellites. The same spectral band is also available from the Flexible Combined Imager (FCI) onboard the Meteosat Third Generation (MTG) satellites. This is the first time of such spectral band from a geostationary satellite imager. In this study, a machine learning (ML) framework is employed to retrieve the TPW of atmosphere during daytime. This ML framework takes in observations of reflectance from bands 0.91µm and 0.86µm, the ratio of reflectance between them, as well as the information of viewing angles as inputs, and uses collocated TPW reanalysis from ERA5 as labels of truth. The methodology and validation will be presented. The application of the 10-minute FCI TPW products to storm nowcasting will be explored. The feasibility of this framework under thin cloudy region will also be presented.