Microwave water vapor atmospheric motion vectors retrieval from polar-orbiting satellites

Atmospheric motion vectors (AMVs) constitute one of the most critical data sources assimilated in numerical weather prediction (NWP), yet current operational wind products fall short to meet forecast requirements. This study addresses a fundamental observational gap in satellite wind retrievals. Traditional polar-orbiting satellite retrievals are limited to high latitudes, and geostationary AMV products are restricted to mid-low latitudes. In the resulting gap regions, only morning-orbit Metop infrared AMVs currently provide limited coverage. This study introduces an optical flow-based atmospheric motion vector retrieval method employing spatiotemporal matching of 183.31 GHz microwave water vapor channel brightness temperatures from NOAA-20/21 Advanced Technology Microwave Sounders (ATMS), enabling highly vertically resolved wind retrievals with clear-sky pixels. Using a fixed 5° × 5° feature tracking regions (5° FTR), the wind speed bias ranges from 0.16 to 0.64 m·s⁻¹, the root mean square error (RMSE) ranges from 3.45 to 3.81 m·s⁻¹, and the wind direction bias was consistently constrained below 27.4°. The overall accuracy achieves the error levels of existing products.

For extremely wind speed conditions, a hybrid-scale FTR optimization model, 3° FTR for slow wind speed region and 10° FTR for those high wind speed region, is also proposed. It can expand the detectable wind speed range from 45 m·s⁻¹ to 70 m·s⁻¹ with a sample size increase of over 10 % per channel. The RMSE for 3° FTR reduces by 0.5 m·s⁻¹, while the 10° FTR achieves a 1.5° reduction in both angular deviations and their standard deviation (STD) at 500 and 450 hPa. For all the hybrid regions in all channels, the RMSE remains within 3.47–3.79 m·s⁻¹, the correlation coefficient is enhanced by about 10 % and the wind direction bias is almost the same as that of the fixed FTR. This hybrid-scale tracking strategy can effectively balance spatial resolution and statistical reliability, and thus provides a new technical paradigm for polar-orbiting microwave AMV retrieval. The resulting afternoon-orbit microwave AMVs deliver a novel wind data source for NWP assimilation systems.