All-sky Assimilation of AMSU-A Window Channels

TitleAll-sky Assimilation of AMSU-A Window Channels
Publication TypeReport
Date Published08/2022
Series/CollectionEUMETSAT/ECMWF Fellowship Programme Research Report
Document Number59
TypeEumetsat Fellowship Programme Research Report
AuthorsDuncan, D, Bormann, N, Geer, AJ
Report Number59

Following the activation of AMSU-A temperature sounding channels in all-sky conditions at ECMWF,
here we consider assimilation of its window channels. These frequencies hold information on clouds and precipitation, columnar water vapour, and low-level winds in 4D-Var, but at ECMWF such channels have previously been assimilated on conically-scanning microwave imagers only. Suitable error modelling is investigated for these channels, with their different sensitivities requiring predictors separate from the sounding channels. In comparison to sounding channels on AMSU-A or similar frequencies on conically-scanning imagers, all-sky error modelling for these channels is more challenging. This is primarily due to their combination of sensitivities to the surface and scattering from hydrometeors aloft in clouds and precipitation that changes with the variable zenith angle.

All-sky assimilation of the AMSU-A window channels is shown to be similar to that of microwave imagers when adequate quality control and observation error modelling are applied; different observation error models are tested, with a 50 GHz-based cloud proxy yielding the best results. The choice of observation error model has a modest effect on the assimilation, though smaller than that of the additional data itself. Experiments using a depleted observing system indicate that active use of these window channels can impart a good fraction of the impact provided by four microwave imagers. When compared to a full observing system, the addition of AMSU-A window channels provides a small but significant improvement in background fits to observations of winds and humidity. These benefits are most notable in the tropics, where the addition of AMSU-A window channels fills in temporal sampling gaps left by the imagers. Signals in the analysis-based verification are small and more mixed, with improvement seen for columnar humidity in the tropics out to day 2 but degradation of low-level temperature forecasts in the southern hemisphere out to day 4. Further testing will be performed to ascertain the suitability of assimilating the AMSU-A window channels in a future operational cycle.

PDF icon Download (11.27 MB)