The Assimilation Of Microwave Humidity Sounder Observations In All-Sky Conditions

Numerical weather prediction (NWP) schemes use a wide variety of satellite observations to help constrain atmospheric analyses. Passive microwave humidity data from channels operating in the water vapour band at 183 GHz, such as those on the Microwave Humidity Sounder (MHS), are an important component of the satellite observing network. At this frequency the observations are very sensitive to scattering from ice crystals and, until recently, quality-control tests were used at the Met Office to exclude assimilation of these data in regions of strong scattering such as thick cirrus clouds. In the work described here we report on improvements to the pre-processing and assimilation of MHS data within the global NWP scheme at the Met Office, with the goal of utilising the data in scenes where liquid and ice cloud strongly affect the radiances. These improvements include the introduction of an observation operator that models the effects of ice scattering, and a scene-dependent observation error model. The error model is based on the estimated cloud ice and cloud liquid in the field of view. The new scheme has increased observation usage of MHS data over ocean by up to 40%. Trials have shown benefits to forecasts in both boreal summer and winter seasons. In the boreal summer period, for example, low-level wind forecast errors have improved in the Northern Hemisphere extratropics by up to 0.6%. Also, our experiments show evidence of significant improvements to the fit between observations and short-range forecasts for humidity-sensitive channels operating in both the microwave and infrared regions of the electromagnetic spectrum.