Consistency Assessment of the Winds in Reanalysis Datasets and the GIIRS Product Using Radiosondes

Accurate three-dimensional (3D) wind products are crucial for weather and climate research. Besides analysis/reanalysis datasets, measurements from the geostationary hyperspectral infrared sounder provide another independent source for obtaining 3D wind profiles at high temporal resolution. In this study, an assessment of the 3D wind product retrieved from Geostationary Interferometric Infrared Sounder (GIIRS) observations, along with several typical reanalysis/analysis datasets from representative centers (e.g., ERA5, CRA-40, MERRA-2, JRA-55 and ART) is conducted, using the independent wind profile observations from the radiosonde network as the truth. It is found that selection of spatial matching thresholds is crucial for objective and reliable evaluation, and homogeneity scene selection aids in reducing inconsistencies between assessed wind data and radiosonde measurements. Reanalysis/analysis datasets show good accuracy between 300 and 900 hPa, but they reveal larger errors near the planetary boundary layer and above 300 hPa. Moreover, wind speed and direction errors are dependent on wind speed, with positive biases dominating at lower wind speeds and shifting towards negative biases as wind speed increases. The ART analysis and ERA5 reanalysis datasets demonstrate the best overall quality, while JRA-55, CRA-40, and MERRA-2 exhibit inferior performance. GIIRS-derived winds perform comparably to other datasets at low wind speeds but show degraded accuracy at higher wind speeds. ART and ERA5 exhibit relatively stable quality across varying weather stability conditions compared to other wind products.