Skip to Main Content
This study introduces a variational approach to retrieve total precipitable water (TPW) over all surface backgrounds including ocean, land, snow, sea-ice, and coastal areas, from microwave sensors. The product has been used routinely by forecasters since its recent operational implementation. The emissivity is accounted for by including its spectrum within the retrieved state vector, which allows for a pixel-to-pixel variation of the emissivity, a factor usually preventing the TPW retrieval over land. The algorithm, implemented operationally at the National Atmospheric and Oceanic Administration (NOAA), is called the Microwave Integrated Retrieval System (MiRS). Its main characteristic, besides its applicability over all surfaces, is its validity under all weather conditions. With a generic design, the algorithm is being applied to the following microwave sensors: (1) AMSU and MHS onboard NOAA-18; (2) NOAA-19 and Metop-A; as well as (3) SSMI/S onboard DMSP-F16 platform. The assessment of the MiRS performances is done by undertaking extensive comparisons to the National Center for Environmental Prediction (NCEP) and the European Centre for Medium-Range Weather Forecasts (ECMWF) analyses, to a network of radiosondes and to existing well-established algorithms over ocean, encompassing a wide variety of meteorological situations. The performance of MiRS TPW is shown to depend on the sensor, the reference data source as well as on the surface background considered. It is shown to behave quite well over all surfaces and in all weather conditions, except when there is rain. Although this study focuses on the retrieval of TPW with an emphasis on non-oceanic surfaces, the underlying application of this study is the potential improvement in the variational data assimilation of Numerical Weather Prediction (NWP) models. Indeed, the same dynamic approach could be employed in order to assimilate more surface-sensitive microwave channels, over a multitude of surfaces.